Building Wave Response Rates in a Longitudinal Survey

Abstract

In national health care surveys, significant amounts of resources are allocated to obtaining the participation of households that constitute the last 5–10% of the overall survey response rate. This study examines the implications of a reduction in targeted first-round response rates in the Medical Expenditure Panel Survey (MEPS), a national health care survey designed to produce estimates of health care utilization, expenditures, and insurance coverage. The potential impact of this modification on annual and overall longitudinal survey response rates, survey estimates, and costs is determined and compared with an alternative approach that subsamples these difficult to field cases. Study findings revealed nonuniform results when evaluating the capacity of these alternative design strategies to achieve precision targets while simultaneously satisfying data collection budget constraints for a variety of medical care expenditure estimates. Based on these results, it was determined that additional experimentation was necessary prior to implementing design modifications.